Testbed and Methods

All today’s testing participants and their only competitor in this test session were tested in two modes: in an open testbed when the mainboard sits horizontally on the desk and the coolers are installed vertically, and in a closed testbed with the mainboard in vertical position.

We put together identical testbeds for our experiments:

• Mainboard: ASUSTek P5K Deluxe/WiFi-AP (Intel P35), LGA 775, BIOS 0809
• Processor: Intel Core 2 Extreme QX9650 (3.0GHz, 1.25V, 2x6MB L2 cache, 4x333MHz FSB, Yorkfield, C0)
• Thermal interface: Arctic Silver 5
• Graphics card: Sparkle GeForce 8600 GT 256MB / 128bit (passive cooling)
• Memory:
  • 2 x 1024MB DDR2 Corsair Dominator TWIN2X2048-9136C5D (1142MHz / 5-5-5-18 / 2.1V);
  • 2 x 1024MB DDR2 CSXO-XAC-1200-2GB-KIT DIABLO (1200MHz / 5-5-5-16 / 2.4V).
• Disk subsystem: Samsung HD501LJ (SATA-II, 500GB storage capacity, 7200rpm, 16MB cache, NCQ)
• Optical drive: Samsung SH-S183L SATA-II DVD RAM & DVD±R/RW & CD±RW
• System case: System case: ASUS ASCOT 6AR2-B Black&Silver (ATX) with 120mm ~900RPM Scythe Minebea fans for air intake and exhaust, and a 120mm ~840RPM Scythe Slip Stream 120 SY1225SL12L fan on a side panel
• Control and monitoring panel: Zalman ZM-MFC2
• Power supply: Enermax Galaxy EGA1000EWL 1000W (a default 135mm ~850RPM fan for intake; 80mm ~1550RPM Noctua fan for air exhaust)

Using the weakest cooling system of our today’s testing participants we managed to overclock our quad-core processor to 3.75GHz with the Vcore increased to 1.475V in the mainboard BIOS. The monitoring utilities reported the core voltage setting a little bit lower than what was set in the mainboard BIOS: around 1.44~1.46V. The system memory was working at 980MHz efficient frequency with 5-5-5-14_2T timings and 2.05V voltage. Later on when we tested the processor for maximum overclocking with each of the coolers, we increased its frequency and voltage in each case.

All tests were performed under Windows XP Professional Edition SP2. SpeedFan 4.34 Beta 44 was used to monitor the temperature of the CPU, reading it directly from the CPU core sensor. We reflashed the mainboard BIOS and updated all the monitoring utilities. As a result, we got the whole bunch of different temperature readings:

As you see, the difference between the readings from different monitoring programs reached 10ºC! It is very hard to determine with software tools which temperature is the closest to reality. That is why we decided not to veer away from our previously established methodology so the result charts will have the readings from SpeedFan utility. To be fair I have to say that the temperature differences between monitoring programs were identical in idle mode and under peak workload. So, no matter what readings I had decided to choose, the difference between tested coolers would have remained the same.

The mainboard’s automatic fan speed management feature was disabled for the time of the tests in the mainboard BIOS. The CPU thermal throttling was controlled with the new RightMark CPU Clock Utility version 2.35.0 that now supports Intel Core 2 Extreme QX9650 processor.

The CPU was heated up with OCCT (OverClock Checking Tool) version 2.0.0 in a 23-minute test with maximum CPU utilization, during which the system remained idle in the first and last 4 minutes of the test:

I performed at least two cycles of tests and waited for approximately 20 minutes for the temperature inside the system case to stabilize during each test cycle. The stabilization period in an open testbed took about half the time. The maximum temperature of the hottest CPU core of the four in the two test cycles was considered the final result (if the difference was no bigger than 1°C – otherwise the test was performed at least once again). Despite the stabilization period, the result of the second test cycle was usually 0.5-1°C higher.

The ambient temperature was checked next to the system case with an electronic thermometer that allows monitoring the temperature changes over the past 6 hours. During our test session room temperatures varied between 24.5 ~ 25°C. It is used as a staring point on the diagrams. Note that the fan rotation speeds as shown in the diagrams are the average readings reported by SpeedFan, and not the official claimed fan specifications.

The noise level of each cooler was measured according to our traditional method described in the previous articles with the help of an electronic noise meter – CENTER-321. The subjectively comfortable noise level was considered 34.5dBA and is marked with a dotted line in the diagram. The ambient noise from the system case without the CPU cooler didn’t exceed 33.2dBA when measured at 1m distance.

We believe it is not enough to just compare the coolers with one another that is why we decided to also compare them against Thermalright Ultra-120 eXtreme super-cooler equipped with Scythe Slip Stream 120 120x120x25mm fan rotating at ~1310RPM (according to the monitoring utilities):